feat: Support a wider range of iterables in SchemaBase.to_dict

altair/utils/schemapi.py

@@ -32,6 +32,7 @@
 import jsonschema.exceptions
 import jsonschema.validators
 from packaging.version import Version
+import narwhals.stable.v1 as nw
 
 # This leads to circular imports with the vegalite module. Currently, this works
 # but be aware that when you access it in this script, the vegalite module might
@@ -486,6 +487,14 @@ def _subclasses(cls: type[Any]) -> Iterator[type[Any]]:
             yield cls
 
 
+def _from_array_like(obj: Iterable[Any], /) -> list[Any]:
+    try:
+        ser = nw.from_native(obj, strict=True, series_only=True)
+        return ser.to_list()
+    except TypeError:
+        return list(obj)
+
+
 def _todict(obj: Any, context: dict[str, Any] | None, np_opt: Any, pd_opt: Any) -> Any:
     """Convert an object to a dict representation."""
     if np_opt is not None:
@@ -510,10 +519,16 @@ def _todict(obj: Any, context: dict[str, Any] | None, np_opt: Any, pd_opt: Any)
             for k, v in obj.items()
             if v is not Undefined
         }
-    elif hasattr(obj, "to_dict"):
+    elif (
+        hasattr(obj, "to_dict")
+        and (module_name := obj.__module__)
+        and module_name.startswith("altair")
+    ):
         return obj.to_dict()
     elif pd_opt is not None and isinstance(obj, pd_opt.Timestamp):
         return pd_opt.Timestamp(obj).isoformat()
+    elif _is_iterable(obj, exclude=(str, bytes)):
+        return _todict(_from_array_like(obj), context, np_opt, pd_opt)
     else:
         return obj
 
@@ -1228,6 +1243,12 @@ def _is_list(obj: Any | list[Any]) -> TypeIs[list[Any]]:
     return isinstance(obj, list)
 
 
+def _is_iterable(
+    obj: Any, *, exclude: type | tuple[type, ...] = (str, bytes)
+) -> TypeIs[Iterable[Any]]:
+    return not isinstance(obj, exclude) and isinstance(obj, Iterable)
+
+
 def _passthrough(*args: Any, **kwds: Any) -> Any | dict[str, Any]:
     return args[0] if args else kwds
 

tests/utils/test_schemapi.py

@@ -1,4 +1,5 @@
 # ruff: noqa: W291
+from __future__ import annotations
 import copy
 import io
 import inspect
@@ -7,11 +8,14 @@
 import jsonschema.exceptions
 import pickle
 import warnings
+from collections import deque
+from functools import partial
 
 import numpy as np
 import pandas as pd
 import pytest
 from vega_datasets import data
+import polars as pl
 
 import altair as alt
 from altair import load_schema
@@ -524,23 +528,23 @@ def chart_error_example__wrong_tooltip_type_in_faceted_chart():
     return (
         alt.Chart(pd.DataFrame({"a": [1]}))
         .mark_point()
-        .encode(tooltip=[{"wrong"}])
+        .encode(tooltip=[{5000}])
         .facet()
     )
 
 
 def chart_error_example__wrong_tooltip_type_in_layered_chart():
     # Error: Wrong data type to pass to tooltip
     return alt.layer(
-        alt.Chart().mark_point().encode(tooltip=[{"wrong"}]),
+        alt.Chart().mark_point().encode(tooltip=[{5000}]),
     )
 
 
 def chart_error_example__two_errors_in_layered_chart():
     # Error 1: Wrong data type to pass to tooltip
     # Error 2: `Color` has no parameter named 'invalidArgument'
     return alt.layer(
-        alt.Chart().mark_point().encode(tooltip=[{"wrong"}]),
+        alt.Chart().mark_point().encode(tooltip=[{5000}]),
         alt.Chart().mark_line().encode(alt.Color(invalidArgument="unknown")),
     )
 
@@ -656,21 +660,21 @@ def chart_error_example__four_errors():
         (
             chart_error_example__wrong_tooltip_type_in_faceted_chart,
             inspect.cleandoc(
-                r"""'{'wrong'}' is an invalid value for `field`. Valid values are of type 'string' or 'object'.$"""
+                r"""'\[5000\]' is an invalid value for `field`. Valid values are of type 'string' or 'object'.$"""
             ),
         ),
         (
             chart_error_example__wrong_tooltip_type_in_layered_chart,
             inspect.cleandoc(
-                r"""'{'wrong'}' is an invalid value for `field`. Valid values are of type 'string' or 'object'.$"""
+                r"""'\[5000\]' is an invalid value for `field`. Valid values are of type 'string' or 'object'.$"""
             ),
         ),
         (
             chart_error_example__two_errors_in_layered_chart,
             inspect.cleandoc(
                 r"""Multiple errors were found.
 
-                Error 1: '{'wrong'}' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
+                Error 1: '\[5000\]' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
 
                 Error 2: `Color` has no parameter named 'invalidArgument'
 
@@ -687,7 +691,7 @@ def chart_error_example__four_errors():
             inspect.cleandoc(
                 r"""Multiple errors were found.
 
-                Error 1: '{'wrong'}' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
+                Error 1: '\[5000\]' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
 
                 Error 2: '4' is an invalid value for `bandPosition`. Valid values are of type 'number'.$"""
             ),
@@ -697,7 +701,7 @@ def chart_error_example__four_errors():
             inspect.cleandoc(
                 r"""Multiple errors were found.
 
-                Error 1: '{'wrong'}' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
+                Error 1: '\[5000\]' is an invalid value for `field`. Valid values are of type 'string' or 'object'.
 
                 Error 2: `Color` has no parameter named 'invalidArgument'
 
@@ -948,3 +952,37 @@ def test_to_dict_expand_mark_spec():
     chart = alt.Chart().mark_bar()
     assert chart.to_dict()["mark"] == {"type": "bar"}
     assert chart.mark == "bar"
+
+
+@pytest.mark.parametrize(
+    "expected",
+    [list("cdfabe"), [0, 3, 4, 5, 8]],
+)
+@pytest.mark.parametrize(
+    "tp",
+    [
+        tuple,
+        list,
+        deque,
+        pl.Series,
+        pd.Series,
+        pd.Index,
+        pd.Categorical,
+        pd.CategoricalIndex,
+        np.array,
+    ],
+)
+def test_to_dict_iterables(tp, expected) -> None:
+    x_dict = alt.X("x:N", sort=tp(expected)).to_dict()
+    actual = x_dict["sort"]  # type: ignore
+    assert actual == expected
+
+
+@pytest.mark.parametrize(
+    "tp", [range, np.arange, partial(pl.int_range, eager=True), pd.RangeIndex]
+)
+def test_to_dict_range(tp) -> None:
+    expected = [0, 1, 2, 3, 4]
+    x_dict = alt.X("x:O", sort=tp(0, 5)).to_dict()
+    actual = x_dict["sort"]  # type: ignore
+    assert actual == expected

tools/schemapi/schemapi.py

@@ -30,6 +30,7 @@
 import jsonschema.exceptions
 import jsonschema.validators
 from packaging.version import Version
+import narwhals.stable.v1 as nw
 
 # This leads to circular imports with the vegalite module. Currently, this works
 # but be aware that when you access it in this script, the vegalite module might
@@ -484,6 +485,14 @@ def _subclasses(cls: type[Any]) -> Iterator[type[Any]]:
             yield cls
 
 
+def _from_array_like(obj: Iterable[Any], /) -> list[Any]:
+    try:
+        ser = nw.from_native(obj, strict=True, series_only=True)
+        return ser.to_list()
+    except TypeError:
+        return list(obj)
+
+
 def _todict(obj: Any, context: dict[str, Any] | None, np_opt: Any, pd_opt: Any) -> Any:
     """Convert an object to a dict representation."""
     if np_opt is not None:
@@ -508,10 +517,16 @@ def _todict(obj: Any, context: dict[str, Any] | None, np_opt: Any, pd_opt: Any)
             for k, v in obj.items()
             if v is not Undefined
         }
-    elif hasattr(obj, "to_dict"):
+    elif (
+        hasattr(obj, "to_dict")
+        and (module_name := obj.__module__)
+        and module_name.startswith("altair")
+    ):
         return obj.to_dict()
     elif pd_opt is not None and isinstance(obj, pd_opt.Timestamp):
         return pd_opt.Timestamp(obj).isoformat()
+    elif _is_iterable(obj, exclude=(str, bytes)):
+        return _todict(_from_array_like(obj), context, np_opt, pd_opt)
     else:
         return obj
 
@@ -1226,6 +1241,12 @@ def _is_list(obj: Any | list[Any]) -> TypeIs[list[Any]]:
     return isinstance(obj, list)
 
 
+def _is_iterable(
+    obj: Any, *, exclude: type | tuple[type, ...] = (str, bytes)
+) -> TypeIs[Iterable[Any]]:
+    return not isinstance(obj, exclude) and isinstance(obj, Iterable)
+
+
 def _passthrough(*args: Any, **kwds: Any) -> Any | dict[str, Any]:
     return args[0] if args else kwds